Inferential Statistics - consist of techniques that allow us to study samples and then make
generalizations about the population from which they were presented.
Sampling error - is the naturally occurring discrepancy or error that exists between a sample statistic
and the corresponding population parameter.
Scales of Measurement
Nominal Scale - consists of a set of categories that have different units. Measurements on a nominal
scale label and categorize observations, but do not make any quantitative distinctions between
observations
Ordinal Scale - consists of a set of categories that are organized in an ordered sequence.
Measurements on an ordinal scale rank observations in terms of size or magnitude.
Internal Scale - consists of ordered categories that are all intervals of exactly the same size. Equal
differences between numbers on scale reflect equal differences in magnitude. However, the zero
point on an interval scale is arbitrary and does not indicate a zero amount of the variable being
measured.
Ratio Scale - is an interval scale with the additional feature of an absolute zero point. With a ratio
scale, ratios of numbers do reflect ratios of magnitude.
Nominal - Magnitude = X, Equal Interval = X, True zero value = X
Ordinal - Magnitude = ✓, Equal Interval = X, True zero value = X
Interval - Magnitude = ✓, Equal Interval = ✓, True zero value = X
Ratio - Magnitude = ✓, Equal Interval = ✓, True zero value = ✓
No ratio data in Psychology as there is no true zero value.
Analytical Normal Distribution - Shapiro Wilk } Null Hypothesis Ho
Ho: The data is normally distributed
P value is equal or < 0.05 = the distribution is normally distributed
P value > 0.05 = the distribution is not normally distributed, reject null hypothesis
Example
W = 0.954
N = 20
P = 0.44
Since 0.44 is more than 0.05, it’s normally distributed. Failed to reject null hypothesis
Always report EVERY detail (W, N, p value)
Graphical Normal Distribution
Box Plot - To check if datas are normally distributed
When the median is at the center of the box and whiskers are almost the same length, we can assume
the data us approximately normally distributed/fairly symmetrical
Bar Plot - Categorical Variable
Descriptie - Skewness & Kurtosis
Skewness - is basically a measure of asymmetry
Kurtosis - is a measure of how “thin” or “fat” the tails of a distribution
What you want is a value closer to 0
Research Hypothesis - involves making a substantive, testable, scientific claim. (Ex. Listening to music
caues people to experience more intense emotions because it arouses many brain areas.)
Statistical Hypothesis - must be mathematically precise. (Ex. Music has a significant effect on the
emotional intensity perceived by listeners.)
Mediates
Differences
Relationships
Effects
Impacts
Predicts
H0 = Null Hypothesis
Ha/H1 = Alternative Hypothesis
Reject the Null
Rejection
Ex. Null Hypothesis = Music does not have a significant effect on emotional intensity perceived by
listeners
Ex. Alternative Hypothesis = Music has a significant effect on the emotional intensity perceived by
listeners.
Statistical Errors - While we ideally would like to always make correct decisions, sometimes we might
commit errors in decision making.
Type I Error - occurs if we reject a null hypothesis that is really true. AKA False Positive
Type II Error - occurs if we retain the null hypothesis when it is in fact false. False Negative
Non-directional test - happens when the alternative hypothesis covers the area on both “sides” of the
null hypothesis, and as a consequence the critical region of the test covers both tails of the sampling
distribution. This produces a two-tailed hypothesis. (X does not significantly affect Y)
Directional test - produces a one-tailed hypothesis ( X is greater than Y) l(X is less than or equal to Y)
Linear Regression
- Is a statistical analysis technique that attempts to model the relationship between two
variables by fitting a linear equation to observed data.
One variable serves as an explanatory variable while the other is the dependent
variable
Method of least-squares - most common method for fitting a regression line
correlation - prediction - causation
Pearson - Linearity
Spearman - Monotonic
Residual - From estimated value
Error - From actual data
Little to no multicollinearity
Multicollinearity - when the independent variables are too highly correlated with each
other. Is concerned with different variables altogether.
Autocorrelation - also known as lagged or serial correlation, this is when residuals are
dependent on each other, I.e. when the value of y(x+1) is not independent from the
value of y(x). Is concerned with the same variable at different time points.
y(hat) = bx + a
H0 : b = 0 (no regression slope)
H1 : b =/ 0 (sig. Regression slope)
Covariates - are simply the continuous data variables
Factors - include your nominal variables
Durbin-Watson autocorrelation test - test of violation
Collinearity statistics checks if the iv are too highly correlated with each other. You
want to ensure that there is no multicollinearity.
Cook’s distance - should have no influential outliers (no more than 1.00)
Model fit checks
R - proviedes you with the linear correlation coefficient of a sample. In essence, it
mirrors the magnitude of a linear relationship as found in a Pearson’s r correlation
(though it may not always reflect the same direction.)
R(2 squared) - coefficient of determination. This is the proportion of the variation in
the DV that is explained by the predictor variable/s. On a simple linear regression, this
is a useful coefficient.
Fear of Death = -0.667 (Life Fulfillment Estimate) + 8.624 (Intercept of Y Estimate)
Simple linear regression was done to test the extent to which participants life
fulfillment predicted their fear of death. Overall, the regression model was statisically
significant, R(squared) = .834, F(1,5) = 25.2, P = .004. Life fulfillment significantly
predicted fear of death, b = -0.667, p = 0.004. The fitted equation was: fear of death =
-0.667 (life fulfillment) + 8.624 (estimate of intercept).
Simultaneous Regression: Examine the unique effects of each predictor on a criterion
at the same time.
Stepwise Regression: Identify (a) how much total variance a set of predictors can
account for in a criterion variable and (b) which predictors can explain more variance
in the criterion. Algorithms determine stepwise sequencing.
Hierarchical Regression: Test the effects of a specific order of predictors as
determined by priori theory or hypothesis. In the hierarchical regression equation, the
researchers set the specific order of predictors.
Adjusted R(squared) is for multiple regressions
AIC (Akaike information criterion) - estimation of prediction error
BIC (Bayesian information criterion) - criterion for model selection.
RMSE (root mean squared error) - measures the avg diff between the values predicted
by a model and the actual cases.
Greather than 1 = sama number before decimal point
Simple linear regression was done to test the extent to which participants'
life fulfillment predicted their fear of death. Overall, the regression model was
statistically significant, R2 = .834, F(1,5) = 25.2, p = .004. Life fulfillment
significantly predicted fear of death, b = -0.667, p = .004. The fitted equation
was: fear of death = -0.667(life fulfillment) + 8.624.
Multiple linear regression was conducted to predicted participants' fear of death based
on their reported life fulfillment and spirituality. Overall, the regression model was
statistically significant, adj. R2 = .929, F(2,4) = 40.1, p= .002. Life fulfillment did not
significantly predict fear of death, b = -0.208, p = 279. However, spirituality
significantly predicted fear of death, b (estimate) = 0.638, p = .034. The fitted
regression model was equivalent to: fear of death = -0.208(life fulfillment) +
0.628(spirituality) + 2.377.
A Pearson product-moment correlation was run to determine the relationship between
height and distance jumped in a long jump. The findings suggest a statistically
significant result. this indicates that there is a strong correlation between height and
distance jumped, (r = .706, p = .005).