Rumor Scale Development

Joshua Chiroma Gandi, PhD

1
Department of General and Applied Psychology, University of Jos, Jos, Plateau State, Nigeria
2
The Psychometric Laboratories, P. O. Box 1550, Jos, 930150, Plateau State, Nigeria
jcgandi2000@[Link]; psychometriclabs@[Link]; gandij@[Link]

Abstract

Rumor refers to unsubstantiated story or information being circulated. Although the more the integrity of

the source implies the more the reliability of rumor, not all that seems reliable would be adjudged as valid. There has

been cogent need for rumor validity assessment, but dearth of construct-relevant scale hampers empirical data

collection. Considering that psychological scales are indispensable for assessment, the present study developed a

suitable and psychometrically sound scale, using cross-sectional design and 570 randomly sampled participants. The

psychometric properties are based on reliability and validity. Reliability (ά=0.78) was determined by item-total

statistics while validity was based on content validity indexes, principal component analysis and the compatibility of

factor model to the data. Seven extracted factors accounted for 92% of the total scale variance. Rumor intensity

score (R=80) corroborated the scale suitability. However, although the newly developed 50-item Rumor Scale is

suitable for adaptation among different populations at various settings, there is need for confirmatory factor analysis

(CFA) which was not implemented in the initial scale development study. Further validations, suggested to include

cross-cultural and trans-national adaptations using CFA and other competing analysis models, can help to establish

sufficient norms.

Keywords: Validity; Scale development; Rumor scale; Gandi psychometric model; Construct-relevant scale

1 Introduction

1.1 Background to the Study

We often hear people say “it is a rumor”, “stop spreading rumors” etc. Conceptually, rumor refers to any

unverified information being circulated among people to make sense of an unclear situation or to manage any threat

or potential threats (Matsumoto, 2009). Rumor, according to Allport and Postman (1947), Matsumoto (2009) and

Rosnow (1991), aims at making sense of an ambiguous situation or even managing physical and psychological

threats. The different types of rumors are grouped based on subject matter as well as the rumor public and their

1
object of collective concern (Matsumoto, 2009). The definition of choice, as used in the present study, refers to

rumor as an unsubstantiated story or information being circulated consciously and or unconsciously based on

significant intensity (Gandi, Ibadin, Musa, Gana, & Saleh, 2018).

1.2 Statement of Problem and Purpose of the Study

Rumor is capable of affecting individuals or even societal values positively or negatively (Willis, 2015).

Although the more the integrity of rumor source the more the perception of its reliability, not everything that is

reliable would be adjudged as valid (Cabera-Nguyen, 2010; Rao & Sinharay, 2007). There is cogent need for rumor

validity empirical assessment but dearth of ‘construct-relevant’ and suitable scale hampers it. Achieving such a need

requires a more optimal model, not a nomenclatural misnomer, to effectively drive the development of a suitable

‘construct-relevant’ scale. The study, which developed suitable scale for rumor validity assessment, adopted Gandi

Psychometric Model (Gandi, 2018) on which the scale’s conceptual framework was designed and operationalized.

1.3 Conceptual Framework Based on Gandi Psychometric Model

The conceptual framework, in this case (Figure 1), answered the “how and why” questions. It sets out how

main stages through which the scale development process moves, i.e. from left to right. It also reflects systematic

sequence of the process, from test conceptualization to scale validation, including “why” particular stage(s) or

variable(s) presides or succeeds the other (Gandi, 2018).

T
E Literature
S Search and
T Reviews
S
C Item C
O Generation A
N Scoring L
C Model E
E Focus
P Group V
T Discussion Scaling A
U Method Test Item Test L
A I
L Tryout Analysis Revision D
I A
Z Expert Reviews T
A I
Item O
T
Pretesting
I N
O
N Cognitive Testing
Interviews

Figure [Link] Development Framework based on Gandi Psychometric Model

2
2 Methods

2.1 Research Design and Study Setting

A cross-sectional design, considered suitable for scale development (Rivers, 2018), was adopted. It

captures specific point in time, proves/disproves assumptions, applies to various types of research, and can create

new theories for in-depth research. The study was conducted at three settings which include a higher institution, a

public organization, and a community neighborhood setting within Jos, the capital city of Plateau State, Nigeria.

Research confederates and conditions within the study setting helped to not only gauge but to also adequately

prevent and avoid any perceived social desirability or other unwanted influence(s) that could amount to raping the

psychometric quality of the scale in any way.

2.2 Target Population and Sample Participants

The study target population consists of members of staff at the selected higher institution and public

organization as well as adult members of the selected community. They included males and females with diverse

religious affiliations, ethnic backgrounds, and professional/occupational statuses. The sample size was determined

by applying the principle of person-to-item ratio (i.e. subject-variable ratio). Considering the lowest and highest

ratios ever used (4:1 and 36:1) respectively and the requirement of using not less than 10:1 for scale pilot-testing

(Rosnow, 1991; Gregory, 2000; Coaley, 2010), the study used person-to-item ratio of 15:1. Since the scale has 50

items, the determined pilot sample size was 750 (i.e. 15x50). However, only 570, including 52% male (n=296) and

48% female (n=274), actually participated in the pilot study. The participants were selected by a stratified random

sampling method. This method helped to ensure adequate representativeness by using stratification factors which

include gender, religion, marital status, employment types, and occupation in the sampling process.

2.3 Materials and Procedure

Materials. The required materials for implementing a comprehensive process of scale development include relevant

instruments and conditions in each case. Materials used as relevant instruments include informed consent forms,

interview schedule forms, demographic data forms, focus group discussions checklist, expert reviews rating rubrics,

and cognitive testing feedback sheets. Others were video camera, writing materials, SPSS software, and the

processing and analysis system (computer). The materials used as conditions include the basic necessary and

sufficient conditions as well as a great deal of miscellaneous (Mackie, 1965) inus conditions, i.e. “insufficient but

non-redundant part of unnecessary but sufficient conditions (inus conditions)”, based on need.

3
Procedure. Scale development process considered the purpose and type of scale as well as suitable psychometric

perspectives (Cohen, Swerdlik & Sturman, 2013; Colton & Covert, 2007; DeVellis, 2012). It was based on nine

stages of the Gandi psychometric model which include test conceptualization, item generation, scaling methods,

item pretesting, scoring models, test tryout, item analysis, test revisions, and scale validation (Gandi, 2018).

Test Conceptualization. Conceiving an idea of the focal construct and what differentiates the proposed scale from

others was cogent. Gregory (2000) and Cripps (2017) suggest that initial thoughts (or self-talk), such as “there ought

to be a test designed to measure ….”, which led to creative and empirical conceptualization of the Rumor Scale for

measuring rumor validity. Confronted questions, as Cohen et al. (2013) observed, include “what is the objective,

content, and format of the Rumor Scale?” Thinking-aloud-questions of this kind paved the way towards reviewing

rumor literature which lends credence to the scale objective, content and format. Rumor was therefore

conceptualized as an unsubstantiated story or information being circulated consciously and or unconsciously based

on significant intensity (Gandi et al., 2018). Equation 1, based on the formula by Allport and Postman (1947),

determines rumor intensity as a mathematical function:

R̴ ixa (1)

where “R” is the rumor intensity which includes the reach and duration of the rumor, “i” is the importance of the

information in the rumor (to hearers or readers), and “a” is the ambiguity (or uncertainty) associated with the rumor

and pictures the persistence of particular pieces of gossip. Rumor intensity depends on the multiplicative

relationship between its importance and ambiguity (see equation 1).

Item Generation. Initial 38 items were deductively derived by conducting literature search and reviews (Figure 1) as

relates to the focal construct (which is “rumor”). This process also provided further useful insights on the type of

statements readily comprehensible across the diverse study populations. Additional 36 items were also inductively

devised by conducting focus group discussions (FGD) with a cross-section of the target population members

referred to as FGD panelists (Colton & Covert, 2007; Gandi, 2018). The 74 items were appropriately worded in line

with factors that constitute the term “rumor” and systematically formatted, by logical question ordering, for

suitability. Item selections were done based on the item content validity index derivable from content validity ratio

(CVR) of the subsequent retained items in each case. The CVR for each item are calculated using the computation

formula by Lawshe (1975) as follows:

4
 𝑁
𝑛𝑒−
2
𝐶𝑉𝑅 = 𝑁 (2)
2

where: ne is the number of FGD panelists who endorsed “essential”, N is the total number of entire FGD panelists

who participated, and 2 is a constant. Furthermore, content validity index (CVI) is computed by finding the

statistical mean of CVR values of the entire scale items as follows:

𝑠𝑢𝑚 𝑜𝑓 𝐶𝑉𝑅
CVI = (3)
𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦

Scaling Methods. A suitable scaling method is a necessity for newly developed scale (Rao & Sinharay, 2007). The

optimally suitable method fall within the category of selective response multiple choice scaling, typically the 5-point

Likert scale type (Likert, 1932; Worthington & Whittaker, 2006). Response options for the scale include ‘strongly

disagree, disagree, not sure, agree and strongly agree. The advantages of and rationale for choosing Likert scale

include ease of scoring, high compatibility for assessment, convenience to participants, and a robust analysis that

would lead to appropriate interpretations.

Item Pretesting. 74 items earlier generated items were rigorously pretested by subjecting them to expert reviews and

cognitive testing interviews (Willis, 2015). The reviewers include subject matter expert (1), content expert (1),

methodology expert (1), and target population experts (2). As scholars including Rao and Sinharay (2007) and

Ojedokun (2016) emphasized, “experts were asked to identify items that they particularly view as important by

indicating whether such items are essential, good, or poor as relates to the focal construct”. The five expert

reviewers rated each item, considering the relevance and respective match (suitability) to the construct, and their

reviews reduced the 74 items to 53 items. The 53 items were further subjected to cognitive testing interviews (CTI)

which used 19 participants (M = male and F = female). They include representative sample of four academic staff

(M=2, F=2), seven senior administrative staff (M=4, F=3), two clinical staff (M=1, F=1), five field staff (M=3,

F=2), and a social worker (F=1). After the survey completion, participants responded to a 7-item “cognitive testing

follow-up sheet”, in their own words. Three items were dropped while 50 were retained and considered for the next

but one stage (i.e. pilot study or test tryout).

Scoring Model. Suitability and ease of scoring are unavoidable features that strengthen a good rating scale. Hence,

the scoring model adopted for this study was a cumulative scoring, being the very aspect which demonstrated that

“the higher the score on the test the higher the test-taker is on the ability/trait/other characteristic. Scoring models

5
must agree with corresponding methods of scaling (Mackie, 1965). That was why the “considered advantages” of

and rationale for choosing Likert scaling method included ease of scoring, high compatibility for assessment,

convenience to participants, and a robust analysis that would lead to appropriate interpretations.

Test Tryout. The scale was tried out (Figure 1) by conducting a pilot study using 570 persons, including 52% male

(n=296) and 48% female (n=274), as participants. Consent forms in which the participants demonstrated voluntary

willingness to participate by appending signatures were provided. The form(s) explain the study need and the

participants’ expectation while issues of their confidentialities and privacies were guaranteed. The demographic

form, which consisted eight items, was administered alongside the newly developed scale to gather descriptive

information on gender, age, marital status, religion, highest qualification, current employers, employment type, and

department/unit. The Rumor Scale items appear self-validating as they helped to control any tendency for responses

to reflect what could be presumed as being a desired effect (i.e. social desirability bias).

Item Analysis. The study maintained cogent culture of ascertaining item reliability index, difficulty index,

discrimination index and validity index which ensured soundness of the scale. Item difficulty index and

discrimination index were qualitatively determined by employing the stated rigorous processes of item pretesting.

The statistical analysis, i.e. quantitative method, was used for reliability index and validity index of the retained

items. Overall, the analyses conducted include content validity index (CVI), item-total statistics, exploratory factor

analysis (EFA), and principal component analysis (PCA).

Test Revisions. Following the pilot study item analyses, the resulting outcomes were reviewed and classified into

two categories. Those items (n=50) found suitable and significant (labelled as category 1) were retained while those

neither sound nor significant (labelled as category 2) were deleted (discarded).

3 Results

The results of examining the content validity, based on equation 3 formula, showed CVIs of 0.66, 0.69 and

0.72 (i.e. CVIs>60) for item generation, expert review item pretesting and cognitive interviews item pretesting. The

CVI of 0.66 resulted from the sample size of focus group discussion (FGD) panelists (n=18) which represent

frequency score in equation 3. The FGD (18) and the sum of CVR (12) therefore determined CVI to 0.66. Item

pretesting, using expert reviewers (5) and the sum of their endorsements CVR (3.45) determined the expert reviews

content validity, CVI to 0.69. The 53 items with content validity of 0.69 (out of the 74 items) were retained while 21

items with content validity below the adopted cutoff point of 0.60 were dropped. Again, item pretesting with

6
cognitive testing participants (n=19) and the sum of their item endorsements CVR (26.4) determined the cognitive

testing content validity, CVI to 0.72. Based on the cognitive testing content validity (0.72), three of the items

(CVI<0.60) were dropped while 50 items (CVI>0.60) were retained. Overall, content validity results shows 0.66,

0.69 and 0.72 based on item generation, expert review pretesting and cognitive testing interview analyses.

Results of rumor intensity (R) computation of the data generated with the newly developed scale, based on

equation 1 formula, shows R = 0.8 (i.e. decimal of R = 80). This is because the rumor importance (i) score was 10

while its ambiguity (a) score was 8. Just as equation 1 noted that R ̴ i x a, which means rumor intensity is the

product of its importance to the hearer (or reader) and its corresponding ambiguity, having the importance (i) as 10

and the ambiguity (a) as 8 implies that the intensity (R) is 80. It must therefore be noted that rumor intensity was

computed according to percentage model and the product (R=80) is converted to a decimal (i.e. R = 80 = 0.8).

The 50 retained scale items, resulting from significant CVIs and construct-relevant rumor intensity (R),

explored had significant minimum mean (2.50) and maximum mean (4.67) as well as the lowest standard deviation

and highest standard deviation (0.98 and 1.81 respectively). None of the retained items had absolute values of more

than 0.30 for skewness and or kurtosis.

Internal consistency reliability based on Cronbach’s alpha was ά = 0.78 and the range of item correlation

coefficients, based on Pearson, r = 0.45 – 0.71 (p<0.01). The results, based on item-total statistics, indicated that all

the items have significant reliability. The analysis, as presented in Table 1, helped to check for items that were

inconsistent with the average behavior of other items on the scale. The reliability (if item deleted) was based on

scale mean, scale variance, corrected item-total correlation, and Cronbach’s alpha (Table 1). Scale mean refers to if

item 1 (for instance) was removed and the remaining items on the scale summed up for all 570 cases then 23.10

would be the mean of the summated items. Scale variance shows that if item 1 (for instance) was removed and the

remaining items on the scale summed up for all 570 cases, then 23.02 would be the variance of the summated items.

In case of the corrected item-total correlation, the correlation between (for instance) item 1and the summated score

was 0.71, as presented in the third column of Table 1. Cronbach’s alpha refers to if item 1 (for instance) was

removed the Cronbach’s alpha reliability coefficient would be 0.7744. By investigating total correlations (see Table

1), only few items have been deleted for failure to satisfy the requirements to be retained due to their low

correlations while 50 items that satisfied such requirements are retained. This further improved the scale reliability

which corroborated the internal consistency reliability (0.78).

7
Table 1. Item-Total Statistics (Showing “if item deleted”)
Scale Scale Corrected Cronbach
Mean Variance Item-Total Alpha (ά)
Correlation
1 Uncertainty about a situation leads to rumor transmission. 23.10 23.02 0.71 .67
2 Rumor keeps one alert and helps to avoid danger or insecurity. 23.68 21.13 0.60 .71
3 Lack of evidence is a key defining feature of rumor. 23.56 22.52 0.54 .68
A rumor may be like “I heard that the real reason those in charge were forced to step down include
4
certain controversial comments they made”. 23.13 23.10 0.61 .69
5 Rumor refers to unverified information being circulated. 23.51 20.49 0.48 .77
6 Information from a credible source is more believable than those spread by non-credible sources. 23.60 22.12 0.45 .73
7 Reducing uncertainty or anxiety helps in managing/preventing rumors. 23.06 21.45 0.66 .76
8 Information from an uncertain source destabilizes my decision making. 23.22 22.04 0.59 .70
9 Rumor affects performance positively. 23.17 20.23 0.68 .63
10 Rumor is an attempt to make sense of unclear situation or to manage threats. 23.10 23.02 0.71 .71
11 Rumor is different from gossip in every sense. 23.68 21.13 0.60 .69
12 It is okay to share any information even without evidence supporting them. 23.56 22.52 0.54 .67
13 Memory limits can reduce rumor transmission. 23.13 23.10 0.61 .68
14 I will not share information that seems not to have any evidence. 23.51 20.49 0.48 .70
15 Uncertain events are believed to be sources of anxiety. 23.60 22.12 0.45 .74
16 Rumors in agreement with my current attitudes are more believable than those that disagree. 23.06 21.45 0.66 .66
17 Reducing belief in a rumor will contribute to managing it. 23.22 22.04 0.59 .71
Rumor may read “it is believe that the main opposition party is behind the ongoing uncompromised
18
industrial disharmony against government”. 23.13 23.10 0.61 .72
19 Rumor affects performance negatively. 23.51 20.49 0.48 .68
20 Anxiety promotes rumor among individuals. 23.60 22.12 0.45 .68
21 Any news that turns out to become a lie, after verification, is a rumor. 23.06 21.45 0.66 .67
22 Information from unreliable source has no guarantee in life. 23.22 22.04 0.59 .72
23 Uncertain information constitutes more uneasiness or significant stress. 23.17 20.23 0.68 .62
A rumor may read “have you heard? … that the businessman in town is rumored to be the financier of
24
terrorists”. 23.10 23.02 0.71 .67
25 Ambitious anticipation is a fertile ground for rumor to thrive. 23.13 23.10 0.61 .64
26 A rumor being circulated repeatedly is more believable to me than those heard just once. 23.51 20.49 0.48 .67
27 Rumor spreads like news if it is still new or current. 23.60 22.12 0.45 .68
28 Rumor is not like news if it is never verified. 23.06 21.45 0.66 .72
29 Rumor can damage an individual’s reputation. 23.22 22.04 0.59 .71
30 Rumor may endear the speaker to the listener. 23.17 20.23 0.68 .69
31 Rumor becomes a fact whenever it is found to be true. 23.10 23.02 0.71 .69
32 I feel relieved when an ambiguous situation suddenly becomes clear. 23.68 21.13 0.60 .66
33 There is something exciting about being kept in suspense. 23.56 22.52 0.54 .61
34 Attempts to fact finding attracts rumor mongering. 23.13 23.10 0.61 .67
35 A rumor refers to any well-orchestrated propaganda being dished out. 23.51 20.49 0.48 .66
36 Lack of having all the information I need frustrates me. 23.60 22.12 0.45 .69
37 Rumor can damage a company’s reputation. 23.06 21.45 0.66 .70
38 Building trust will contribute to preventing and managing rumors. 23.22 22.04 0.59 .75
39 Rumor is an evaluative chat about individuals which may or may not be verified. 23.13 23.10 0.61 .65
40 Rumor keeps people on their toes for effective prevention or management of threats. 23.51 20.49 0.48 .64
41 A rumor is an idle or slanderous talk about an individual who is not present. 23.60 22.12 0.45 .67
42 Belief system determines the strength of rumor transmission. 23.06 21.45 0.66 .77
43 Social status contributes to rumor transmission. 23.22 22.04 0.59 .75
44 Rumor can foster hatred among persons or groups. 23.17 20.23 0.68 .71
45 A story that brings uncertainty easily disorganizes me. 23.10 23.02 0.71 .67
46 Falsification of certain rumors can reduce my belief in similar brand of rumors. 23.13 23.10 0.61 .64
47 I get worried if any sensitive information I was told remains uncertain. 23.51 20.49 0.48 .67
48 Rumor may read “be informed that mindless terrorists are planning attacks in town next week”. 23.60 22.12 0.45 .68
49 Enhancing appropriate communication can prevent rumors. 23.06 21.45 0.66 .72
50 Rumor may be circulated as “I heard that 45 members of staff will be laid off from this organization”. 23.22 22.04 0.59 .71

Exploratory analysis and principal component analysis extracted seven factors with respective item

membership of 10, 8, 7, 10, 6, 5 and 4 which accounted for 92% of the total scale variance (Table 2). These factors

are factor 1 (rumor diagnostics), factor 2 (rumor transmission), factor 3 (rumor impact/effect), factor 4 (response

attitude), factor 5 (rumor belief role), factor 6 (rumor type), and factor 7 (rumor management). The Rumor

Diagnostics Subscale consists of 10 items (3, 5, 10, 11, 21, 28, 31, 35, 39 and 44), Rumor Transmission Subscale

has 8 items (1, 13, 20, 25, 27, 34, 42 and 43), Rumor Impact Subscale has 7 items (2, 9, 19, 29, 30, 37 and 40),

Rumor Response Attitude Subscale consists 10 items (8, 12, 14, 22, 23, 32, 33, 36, 45 and 47), Rumor Belief

8
Subscale has 6 (6, 15, 16, 26, 41 and 46), Rumor Type Subscale has 5 items (4, 18, 24, 48 and 50), and the Rumor

Management Subscale consists of 4 items (7, 17, 38 and 49).

Table 2. Factor Loadings

Factors
1 2 3 4 5 6 7
1 Uncertainty about a situation leads to rumor transmission. .61
2 Rumor keeps one alert and helps to avoid danger or insecurity. .71
3 Lack of evidence is a key defining feature of rumor. .60
A rumor may be like “I heard that the real reason those in charge were forced to step down include certain
4
controversial comments they made”. .60
5 Rumor refers to unverified information being circulated. .88
6 Information from a credible source is more believable than those spread by non-credible sources. .65
7 Reducing uncertainty or anxiety helps in managing/preventing rumors. .97
8 Information from an uncertain source destabilizes my decision making. .96
9 Rumor affects performance positively. .57
10 Rumor is an attempt to make sense of unclear situation or to manage threats. .76
11 Rumor is different from gossip in every sense. .69
12 It is okay to share any information even without evidence supporting them. .77
13 Memory limits can reduce rumor transmission. .78
14 I will not share information that seems not to have any evidence. .71
15 Uncertain events are believed to be sources of anxiety. .72
16 Rumors in agreement with my current attitudes are more believable than those that disagree. .76
17 Reducing belief in a rumor will contribute to managing it. .74
Rumor may read “it is believe that the main opposition party is behind the ongoing uncompromised .71
18
industrial disharmony against government”.
19 Rumor affects performance negatively. .99
20 Anxiety promotes rumor among individuals. .97
21 Any news that turns out to become a lie, after verification, is a rumor. .75
22 Information from unreliable source has no guarantee in life. .57
23 Uncertain information constitutes more uneasiness or significant stress. .76
A rumor may read “have you heard? … that the businessman in town is rumored to be the financier of
24
terrorists”. .88
25 Ambitious anticipation is a fertile ground for rumor to thrive. .85
26 A rumor being circulated repeatedly is more believable to me than those heard just once. .56
27 Rumor spreads like news if it is still new or current. .95
28 Rumor is not like news if it is never verified. .76
29 Rumor can damage an individual’s reputation. .57
30 Rumor may endear the speaker to the listener. .95
31 Rumor becomes a fact whenever it is found to be true. .60
32 I feel relieved when an ambiguous situation suddenly becomes clear. .91
33 There is something exciting about being kept in suspense. .73
34 Attempts to fact finding attracts rumor mongering. .85
35 A rumor refers to any well-orchestrated propaganda being dished out. .71
36 Lack of having all the information I need frustrates me. .94
37 Rumor can damage a company’s reputation. .64 .63
38 Building trust will contribute to preventing and managing rumors. .94
39 Rumor is an evaluative chat about individuals which may or may not be verified. .83
40 Rumor keeps people on their toes for effective prevention or management of threats. .70
41 A rumor is an idle or slanderous talk about an individual who is not present. .78
42 Belief system determines the strength of rumor transmission. .90
43 Social status contributes to rumor transmission. .82
44 Rumor can foster hatred among persons or groups. .79
45 A story that brings uncertainty easily disorganizes me. .94
46 Falsification of certain rumors can reduce my belief in similar brand of rumors. .95
47 I get worried if any sensitive information I was told remains uncertain. .95
48 Rumor may read “be informed that mindless terrorists are planning attacks in town next week”. .99
49 Enhancing appropriate communication can prevent rumors. .85
Rumor may be circulated as “I heard that 45 members of staff will be laid off from this organization”. .89
50

4 Discussions

Psychological measurements are largely based on the nearest approximation of the phenomenon being

measured while psychometric scales are indispensable tools needed for conducting such measurements. Scale

development researchers, including Morgado, Meireles, Neves, Amaral and Ferreira (2017), however decried

limitations resulting from previous scale developments which must be overcome to ensure more optimal scale. The

present study essentially developed a suitable scale, referred to as the “Rumor Scale”, for an empirical rumor

9
validity assessment based on robust construct-relevant methodology (Gandi et al., 2018). Required optimality was

achieved by systematic and rigorous implementation of the empirically designed Gandi Psychometric Model’s 9-

stage process (Gandi, 2018). This made the Rumor Scale more construct-relevant and psychometrically sound than

other scales not subjected to such psychometric rigors.

Structurally, the study conceptual framework which shows empirical systematism led to answering the

“how” and “why” questions in the process of devising/deriving needed items is of essence. The numerous materials

adopted and used have been helpful in effective generation and efficient tracking of data for evidential

documentation accuracy and analysis precision. Mackie (1974), as corroborated by Gandi et al. (2018), believe that

both necessary and inus conditions used have effectively complemented those material instruments. Working with

strictly representative sample of participants, as earlier observed (Morgado et al., 2017; Tabachnick & Fidell, 2007),

has been rewarding in every sense as it contributes to ensuring that the newly developed Rumor Scale is optimally

construct-relevant (Gandi et al., 2018). The implemented person-to-item ratio (i.e. subject-variable ratio) model at

high significance level have added value to the study sample optimality.

Using PCA does not imply being unmindful of the suggestion, as extended by Cripps (2017), to avoid using

PCA as a precursor to confirmatory factor analysis (CFA). The uniqueness of using PCA at the scale development

stage, in this case, is that the use of CFA was not yet required at this level of the study (Gandi, 2018). Reliability

was established based on findings from item-total statistics which checked items that were inconsistent with the

average behavior of other items. Critical analysis and inspection of the scale mean, scale variance, and Cronbach’s

alpha if item deleted provided more insight which helped to avoid construct-irrelevant items by deleting them.

Pretesting based on cognitive testing interviews, which helped in revising and modifying certain items, has

corroborated the improved suitability of the scale items among target population. For Chi-Square tests for model

goodness of fits, both parallel test and strict parallel test results presented the Rumor Scale as a significant

instrument that it is (Zhu & Liu, 2017). The assumption that the r population means are not equal but all the r factor

loadings and population unique factor variances are equal was supported by the parallel test, while the assumption

that all the r population means, factor loadings, and unique factor variances was supported by strict parallel test. The

significant rumor intensity level elicited by the Rumor Scale, which supported reliability and validity, is a

complementary empirical evidence of adequate suitability for assessing rumor validity. This supports Gandi et al.

(2018) who opine that even as rumor predicts good news, it equally triggers distress. However, any distress resulting

10
from conflicts and or trauma is avoidable if knowledge, skills, other resources and support are appropriately

deployed (Gandi & Wai, 2010). The Rumor Scale elicits data that leads to such useful information.

Empirically, as Coaley (2010) emphasizes, the three major sources of validity evidence are based on the

content issues (content-related validity), the relationships with measures of other variables (criterion-related

validity), and the internal structure of the selection procedure (construct-related validity). However, validity of the

present scale was based more on content validity and construct validity at its development phase. The main

limitation was lack of using confirmatory factor analysis in the present study (CFA). The scale will be further

validated to establish norms by implementing the CFA and other competing analysis models. Further validations,

including cross-cultural and trans-national adaptations, which can sufficiently establish diverse norm(s) are

suggested.

5 References

Allport, G. W., & Postman, L. (1947). The Psychology of Rumor. New York: Holt.

Cabrera-Nguyen, .(2010). Author Guidelines for Reporting Scale Development and Validation Results. Journal of

the Society for Social Work and Research, 1(2), 99-103.

Coaley, K. (2010). An introduction to psychological assessment and psychometrics. London, England: Sage.

Cohen, R. J., Swerdlik, M. E., & Sturman, E. D. (2013). Psychological Testing and Assessment: An introduction

to tests and measurement (8th Ed). New York: McGraw-Hill.

Colton, D., & Covert, R. W. (2007). Designing and Constructing Instruments for Social Research and

Evaluation. London: Wiley.

Cripps, B. (2017). Psychometric Testing: Critical Perspectives. USA: Wiley Blackwell

DeVellis, R. F. (2012). Scale Development Theory and Applications: Applied Social Research Methods Series.

Washington DC: Sage.

Gandi, J. C. (2018). Development and Validation of Health Personnel Perceived Quality of Life Scale (Unpublished

doctoral thesis). University of Ibadan, Nigeria.

Gandi, J. C., Ibadan, J., Musa, B., Gana, J. G., & Saleh, D. A. (2018, November). Perceived Impact of Rumor in

Coexistence among Ethnically Diverse Community. Paper session presented at the Annual Scientific

Conference of Nigerian Psychological Association, Lagos Nigeria.

Gandi, J. C., & Wai, P. S. (2010). Impact of Partnership in Coping in mental heal recovery: An experimental study

11
 at the Federal Neuro-Psychiatric Hospital Kaduna. International Journal of Mental Health Nursing, 19(5),

322-330.

Gregory, R. J. (2000). Psychological Testing: History, Principles, and Applications. Boston: Allyn and Bacon.

Lawshe, C. H. (1975). A Quantitative Approach to Content Validity. Personnel Psychology, 28, 563-575.

Likert, R. (1932). A technique for the measurement of attitudes. Archives of Psychology, 140.

Mackie, J. L. (1965). Causes and Conditions. American Philosophical Quarterly, 12, 245-265.

Mackie, J. L. (1974). The cement of the universe: A study of causation. Oxford England: Oxford University Press.

Matsumoto, D. (2009). The Cambridge Dictionary of Psychology. United Kingdom: Cambridge University Press.

Morgado, F. F. R., Meireles, J. F. F., Neves, C. M., Amaral, A. C. S., & Ferreira, M. E. C. (2017). Scale

development: ten main limitations and recommendations to improve future research practices. Psicologia:

Reflexao E Criyica, 30(3), 1-20.

Ojedokun, O. (2016). Development and Psychometric Evaluation of the Littering Prevention Behavior Scale.

Ecopsychology, 8(2), 138-152.

Rao, C. R., & Sinharay, S. (2007). Handbook of Statistics on Psychometrics, Vol. 26.

Rivers, J. (2018). Cross-sectional Study: Definition, Advantages, Disadvantages & Example. Retrieved from

[Link]

on 22/09/2018

Rosnow, R. L. (1991). Inside rumor: A personal journal. American Psychologist, 46(5), 484.

Tabachnick, B. G., & Fidell, L. S. (2007). Using multivariate statistics. Fifth Edition. Pearson Education.

Tinsley, H. E. A., & Tinsley, D. J. (1987). Uses of factor analysis in counseling psychology research. Journal of

Counseling Psychology, 34, 414-424.

Willis, G. B. (2015). Analysis of the Cognitive Interview in Questionnaire Design. Oxford University Press.

Worthington, R. & Whittaker, T. (2006). Scale Development Research: A content analysis and recommendation for

best practices. Counselling Psychologists, 34, 806-838.

Zhu, X., & Liu, F. (2017). Research on Behavior Model of Rumor Based on Systems Dynamics. Hindawi

Complexity, 2, 1-9.

12