Estimation Assignment

Pablo Valdez

Helms School of Government, Liberty University

CJUS 745: Quantitative Methods of Research

Dr. Marc Weiss

February 22, 2026

The confidence interval is believed by the many statisticians as one of the most

basic understandings in statistical inference that can provide researchers the usage of

sample data in their population parameter estimation (Smithson, 2003). The underlying

theoretical assumption was based on the frequentist concept of probability, which

defined probability as the long run relative frequency of an event which may be modeled

by an idealized infinite series of replications, then make an inference about the sample

statistics and the unobservable population parameters in a probabilistic way, a fixed, but

unknown constant.

Theoretical Foundation and Interpretation

Interpreting confidence intervals involves understanding the difference between

statements of probability before or after sampling. Before sampling there exists a

probability statement because the end points are random variables. With a sample of

data, they are no longer random, and the confidence interval can be narrowed down to

an actual 95% confidence interval. It is this step that is frequently misunderstood, with

researchers believing the population probability is 95% that it will lie in the interval. It is

the probability that the procedure used to construct the C.I. will produce an interval that

contains the true parameter in many repetitions. For example, if 100 experiments are

conducted, then 95 of the C.I. will contain the true parameter, 5 will not. This is how a

confidence level c is interpreted (Cumming & Finch, 2001; Smithson, 2003).

Normal Distribution and the Central Limit Theorem

samples of size n will tend to be normal if either (a) the population standard deviation is

known, or (b) sample size is large (n>30), regardless of the underlying population

distribution (Dekking et al., 2005). This property, termed the Central Limit Theorem,

allows confidence intervals to be calculated by assuming the sample means follow a

standard normal distribution if either of (a) or (b) are true. For these conditions, the

confidence interval estimate for the population mean is x ± z* (σ / √n) (Dekking et al.,

2005).

Student's t-Distribution for Small Samples

For small sample where population standard deviation is unknown data is

assumed to follows student t distribution, we can find CI from t distribution (Welch,

1939). Student t distribution is a modification of the normal distribution, and it has one

additional source of variation to consider the new uncertainty which comes from an

unstable standard deviation estimate from the sample, to compensate for this there is

an addition of pair of heavier tails to the distribution. The number of degrees of freedom

for t distribution is n − 1 since this is the number of independent information pieces

within the estimate of the variance. Looking at the Banik and Kibria (2024) study

comparing various methods, t distribution approaches normal distribution when samples

size is large but gives a wider more conservative interval for smaller sample sizes and

hence provides higher chance of achieving the nominal confidence level. The previous

formula for a confidence interval with t distribution replaces z and standard deviation

with t*, t critical value and s, sample standard deviation. The formula is x̄ ± t* (s / √n)
4
ESTIMATION ASSIGNMENT
where s is the sample standard deviation and t* is the t critical value with n − 1 degrees

of freedom (Welch, 1939; Banik & Kibria, 2024).

Confidence Intervals for Population Proportions

Confidence intervals for population proportions are calculated using the normal

approximation to the binomial distribution, a valid method when np and n (1 − p) are

sufficiently large (often when > 5 or 10) (Clopper & Pearson, 1934). This normal

approximation permits treating the sampling distribution of the sample proportion as

approximately normally distributed and using a standard normal critical value. The

formula for the population proportion confidence interval is p ± z* √ (p (1 − p)/n) where p

is the sample proportion and a common method used in opinion polling and survey

research to estimate population proportions with a degree of confidence (Clopper &

Pearson, 1934).

Sample Size Determination

A similar application is by estimating the sample size that should be taken to

have the desired margin of error at the desired level of confidence. It is calculated using

the margin of error formula when solved for n = (z*σ / E) ², setting E as the margin of

error desired. For estimations on proportions when no information of population

proportion is available, a safe value is p = 0.5, which overestimates the required sample

size and thus yields enough cases for adequate accuracy (Smithson, 2003).
5
ESTIMATION ASSIGNMENT
Question 2: Interpreting Poll Results and Bias

Public reporting of survey findings usually occurs through point estimates plus

margins of error. For example, an opinion poll of 385 residents of Honolulu reported that

78% favored a mandatory jail term for DWI conviction with margins of error of ±3

percentage points. This leads to a 95% confidence interval from 75% to 81% (Groves et

al., 2009). The confidence interval allows us to identify plausible values of the actual

population proportion (Groves et al., 2009).

Sampling Error versus Systematic Bias

A distinction important in the study of surveys is that between sampling error and

systematic bias, the latter of which is not accounted for by the margin of error (Groves

et al., 2009). Sampling error results from the random variation associated with selecting

a sample and can be decreased by using a larger sample size; systematic bias results

from errors in study design, data collection and measurement and cannot be

compensated by using a larger sample size. The five main types of error in a survey

include sampling error, nonresponse error, response error, measurement variability, and

coverage or frame issues (Groves et al., 2009; Smithson, 2003).

Question wording is a large source of systematic error in survey research.

Questions which highlight certain features of an issue can prime respondents to

consider these issues more heavily when answering the question (Tourangeau et al.,

2000). This means that if, for example, questions on mandatory minimum sentencing

emphasize horrific injuries inflicted upon the public, a higher proportion of the public will
6
ESTIMATION ASSIGNMENT
favor mandatory minimum sentencing than they otherwise would be due to the salience

of public safety interests. Conversely, framing questions of mandatory minimum

sentencing in terms of the problem of prison overcrowding would likely reduce support

as public costs are made more salient. The resultant confidence interval, although

correct in terms of the degree of error from sampling variability, would be biased from

the true population proportion (Tourangeau et al., 2000; Groves et al., 2009).

Precision versus Validity

This difference between precision and validity is critical to the understanding of

what the limitations of the confidence interval are in survey research. Precision refers to

how much a statistic is influenced by random sampling error as represented by the

margin of error or width of the confidence interval. Validity, however, refers to how well

a survey is measuring the thing that it claims to measure and how well a survey's design

yields an unbiased estimate of the population parameters (Cumming & Finch, 2001).

This means that a survey might be very precise very small margin of error but invalid

systematic bias i.e., it may accurately estimate the population value again but be off the

actual true value.

These distinctions are especially important when applied to the area of criminal

justice research in which public opinion survey data are used in formulating policy

decisions for both the enactment and reform of criminal statutes (Groves et al., 2009).

Those policymakers attempting to inform themselves of the public’s feelings about the

criminal justice system using public opinion survey data cannot rely on the survey’s

small confidence interval as proof that the survey is providing a valid assessment of
7
ESTIMATION ASSIGNMENT
public opinion. Poor wording of survey questions, inadequate sampling frames, and/or a

high level of nonresponse bias may produce a significant public opinion survey with

results that may be quite deceiving to the researcher (Tourangeau et al., 2000; Groves

et al., 2009).

Best Practices for Survey Research

The researchers must employ best practice for survey design to come up with

appropriate and accurate estimation of population attitudes. Questions need to be

neutral and nonemotionally, which do not direct the respondent towards certain

responses (Fowler, 2013; Groves et al., 2009). Good sample design techniques

particularly those involving probability samples i.e., each member of the population to

be estimated has a nonzero known probability of being sampled need to be used.

Methodological details e.g., sampling frame, response rate, precise question wording

must be reported so that readers can evaluate bias (Groves et al., 2009).

Conclusion

One very important component of statistical inference is that of a confidence

interval; it provides the researcher with a range which we expect to contain the

population parameter in the long run while simultaneously acknowledging the amount of

uncertainty that can be associated with sample variation. It is vital for any researcher to

comprehend that a confidence interval is not a probability statement about a specific

interval containing a population parameter but instead represents a description of the

performance characteristics of a method of estimation when used repeatedly (Cumming

difference between sampling error and systemic bias and keep in mind that a

confidence interval only controls for sampling error. In the research realm of criminal

justice as in many other fields dealing with applied issues awareness of both statistical

and methodological concerns is necessary for drawing accurate and valid conclusions

about population attributes and informing evidence based public policy (Groves et al.,

2009).

Christian faith also influences the way in which we go about statistical estimation

Christian scholars view statistics as an extension of the creation mandate to use

creation well Genesis 1:28 New International Version while understanding the limits of

human understanding when compared to God's comprehensive knowledge Psalm 147:5

New International Version. The difference between the nature of sampling error and the

nature of bias in our data is similar to that between human finitude and the manmade

failure that is unethical decision-making Proverbs 11:1 New International Version but

accurate weights find favor with him Christians involved in criminal justice research

should strive to be statistically honest and use this as a way to honor God. Ultimately

Christians must search for truth humbly Proverbs 3:5–6 New International Version and

then apply their knowledge and research in such a way that it is an expression of love to

our fellow man Galatians 5:13–14 New International Version statistics and proper

methodology are ways in which we can serve our neighbor (Zondervan, 2011).

References:
9
ESTIMATION ASSIGNMENT
Banik, S., & Kibria, B. M. G. (2024). Comparing confidence intervals for the mean of
symmetric and skewed distributions. Symmetry, 16(11), 1424.
[Link]

Clopper, C. J., & Pearson, E. S. (1934). The use of confidence or fiducial limits
illustrated in the case of the binomial. Biometrika, 26(4), 404–413.
[Link]

Cumming, G., & Finch, S. (2001). A primer on the understanding, use, and calculation
of confidence intervals. Educational and Psychological Measurement, 61(4), 532–574.
[Link]

Dekking, F. M., Kraaikamp, C., Lopuhaä, H. P., & Meester, L. E. (2005). A modern
introduction to probability and statistics Understanding why and how. Springer.
[Link]

Groves, R. M., Fowler, F. J., Couper, M. P., Lepkowski, J. M., Singer, E., &
Tourangeau, R. (2009). Survey methodology 2nd ed. Wiley.

Smithson, M. (2003). Confidence intervals. SAGE Publications.

Tourangeau, R., Rips, L. J., & Rasinski, K. (2000). The psychology of survey response.
Cambridge University Press. [Link]

Zondervan. (2011). Holy Bible New International Version. Zondervan.