Untitled

• The Quantitative Analyses session will revise analyses covered in PSY140.
• Participants will work with data, comparing JASP and SPSS for portfolios.

Data Overview

• In PSY140, observation studies, surveys, and experiments were covered.
• The type of data and focus, whether on differences or relationships, determine what analyses are possible.
• Familiarity with data terms and distinctions is important for conducting the right analysis.
• Raw data needs cleaning (data wrangling) before analysis.
• The kind of study being done (differences or relationships) and the type of data collected dictate the analyses used; this is referred to as "Assumption Checks".
• Data types are classified as continuous or discrete, and assessment of normalcy is a parameter when performing “Assumption Checks”.
• Descriptive and inferential data usage depends on these assumptions.
• Researchers need to generally know the analysis needed before collecting data.

Data Collection

• Before data collection, determining the data level is important
• Data levels include:
• Nominal: for example, favorite ice cream
• Ordinal: for example, ranking of different ice creams
• Interval: for example, rating temperatures of different ice creams
• Ratio: for example, rating likeability of ice creams on a scale of 1-10
• Data level relates to the Dependent Variable (DV)

Personality Example

• Focus is analyzing personality traits (light vs dark triad) linked to engagement in dark tourism.
• Discrete data collection is possible, categorizing participants as higher on the light or dark triad, or as wanting to visit or not.
• Continuous data collection is possible through scoring people on personality traits, or gauging interest in visiting on a scale of 1-10.

Assumption Checks

• Several steps are needed before analysing data to see if it is parametric or non-parametric.
• Ensuring data levels correlate with test assumptions is vital.
• Normality should be considered when checking assumptions
• Normal distribution can be determined based on Skew and Kurtosis analysis.
• Skew is of higher important, however Kurtosis refers to the tails of a distribution.
• Leptokurtic is classified as tall and thin.
• Platykurtic is classified as short and wide.
• Data distribution normality can be checked using histograms or the Shapiro-Wilk test.
• These tests should not take a large focus in research, and are simply checks to carry out.

Descriptive Statistics

• Descriptive statistics describe aspects of observed data (collected from a sample).
• The measures include:
• Mean (M): Average data, adding all and dividing by participation total
• Median (Mdn): Middle data when each “datum” is ascending order
• Mode (Mo): The most common data point
• If mean, median and mode are equal, distribution is normal.
• Distribution gets descriptive statistics.
• Mean and standard deviation (SD) is reported.
• The median and interquartile range (IQR) are reported.
• Assumption checks determine what we report.
• Mode is for nominal data.
• Distribution is dependant if data is skewed.
• Assessing outliers in the dataset is part of the Normality Assumption.
• The mean is affected by outliers; the median is not to the same extent.

Central Tendency

• If data violates the Normality Assumption, reporting the median (and IQR), not the mean (and SD) is recommended.

Inferential Statistics

• While descriptive statistics show our sample, inferences about the population are made using inferential statistics.
• For analysis, specific statistics are collected (e.g. x², r, t); three analyses are upcoming.
• Obtained results require statistically significant assessment
• Significance occurs when the p-value is below the alpha level.
• The alpha level is 0.05 (5%) in psychology (different fields vary).
• The null hypothesis is rejected if the p-value is below 0.05.
• Given observed results(or extreme results), there's under 5% chance of nothing truly going on.
• P-values are calculated for us using JASP and SPSS.
• Alpha level and degrees of freedom are used to work standard p-value.
• Mistakes can still be made with p-values:
• Type I error (alpha): falsely rejecting the null hypothesis.
• Type II error (beta): incorrectly failing to reject the null hypothesis.
• The alpha level denotes making a Type I error, and the study should have power to make it true.
• Increasing the alpha level has greater chances of incorrectly rejecting the null (Type I).
• Decreasing this increases the chance of not rejecting null (Type II).
• Adequate subject count boosts power (also variance and study).
• Value for power should aim to have 80% or 0.8.
• Using effect sizes like Cohen's d is more effective, as P values are affected by low sample counts/participation.

Inferential Statistics Explained

• Data collected is from the sample that is drawn from a population.
• Determining if observed results are by chance (p-level is less than specified alpha)?
• Effect sizes ensure findings validity.
• The main aim is to infer sample mean correlates to population mean.
• Confidence Intervals can be used in order to assess an appropriate amount of values in which the means are likely to fluctuate in the true population.
• Psychology uses 95%; if the study was re-ran close to 100 times, it is expect that 95 of them would show the same population size.

Analysis Overview

• Relationships between 2 variables utilizes Pearson correlation
• Spearman and Kendall-Tau correlations are effective for Non-parametric cases.
• Frequencies of different responses utilizes a Chi-squared test
• Fishers exact test’s test frequencies as well.
• Differences Within Groups:
• Paired T-tests
• Wilcoxon tests.
• Differences between Groups:
• Independent T test
• Mann-Whitney test.

Testing Frequencies

• Whether observed data (collected count) varies from expected count is assessed.
• Non-parametric testing requires assumptions be met in particular: Data must be nominal, with values that do not affect other values, in mutually exclusive columns showing a value of 5 for at least 80% of cells.

Chi Squared Testing

• The Chi-Square test is 2 way, assessing 2 categorical variables.

Pearson's Correlations Testing

• Assessing the relationship between people.
• Assess Spearman’s Rho and Kendall’s Tau if the normalcy assumption is not met.
• There should be a relationship between your variables

T-Tests

• Compare the standardised difference between 2 groups for higher explanation.
• The mean of one level being different to the mean of the other level is the general explanation.
• Utilise paired and independent T-tests.

Software Overview

• Last year JASP was generally used, but it's more common to use the software SPSS in the professional work place.
• JASP and SPSS will be used during every analysis.
• Canvas guides and program comparisons will be presented in the 'Data Analysis' page of participants portfolio's.

Implications

• After determining whether a null hypothesis is rejected, consider the applications and implications of findings.
• Focus potential implications to the study.
• This is the big point when writing the discussion.