HTH 408- Quiz #2

Random error

occurs by chance, can also occur due to social desirability

systematic error

consistent biases affecting a measurement, can be due to poor technique

public health concepts

used to assess the quality of research studies; Indicate how well a method, technique, or test measures that we are interested in measuring.

reliability

describes the degree to which the results can be repeated or replicated under the same conditions

inter observer/rater reliability

measures the degree of agreement between different people observing or assessing the same thing

test retest reliability

obtained by administering the same test twice over a period of time

internal consistency reliability

a way to determine if all questions on a survey, test, or scale are measuring the same thing, most popular method is Cronbach’s alpha

data triangular

research method where researchers use multiple sources of data to examine the same phenomenon

Validity

Ability of data collection tool to accurately capture and measure the construct we are interested in measuring

construct validity

measures the concept, represents the components we are interested in

content validity

fully representative

face validity

appear to be suitable

criterion validity

compared to standard tools, gold standard

internal validity

if the study design accurately measures the intended conflict within the study

external validity

considers how well the results can be generalized to other populations and contexts

Hawthrone effect

people's tendency to behave differently when they become aware that they are being observed.

determine the population of interest

what is the first step to finding the best sampling technique

probability sampling

randomly selected and sampling frame is often used (quantitative studies)

simple random sampling

complete randomization without any specific boundaries

stratified random sampling

dividing a population into groups or strata and then randomly selecting a sample from each group 

Non-probability sampling

participants are NOT randomly selected, sampling frame may not be available (qualitative studies)

Convenience sampling

participation happens by availability

Snowball sampling

second participant finds third and so on, useful when population is hard to reach or studying a rare disease

Purposive sampling

based on specific characteristics relevant, most valuable insights for the study

Quota sampling

compare different groups within the population of interest, researchers need a sample of predetermined quota or proportions

Data analysis

process of systematically collecting, cleaning, transforming, describing, modeling, and interpreting data

Excel, SPSS, SAS, stata, R

Software packages used

categorical variables

ex: gender, favorite movies (1 for male and 2 for female)

ordinal variables

represent an ordered sequence of responses such as strongly agree or disagree

numerical variable

ex: age, income, height (ratio and interval)

Absolute frequency

reflects the number of times an event has been repeated 

Cumulative percent

shows the collective percentages of categories together 

Bar graph

vertical bars represent each variable, qualitative

Histogram

bars are connected, quantitative, vertical bars represent frequency

line graphs

used for depicting two or more categories simultaneously, quantitative

Scatter plots

dots on graph, dependent on vertical independent one horizontal

Pie Charts

always represents 100%

Normal distribution

A perfect theoretical distribution where the mean, mode, and median are equal to one other, ALWAYS REPORT MEAN

central tendency

includes mean, median, mode

asymmetrical data, also better for outlier data

report median for which type of data

standard deviation

estimate for the average distance that scores deviate from the mean

68%

percent of scores that lie within one SD of mean

95%

percent of scores that lie within two SD of mean

99.7%

percent of scores that lie within three SD of mean

correlation

approximate a linear relationship

regression

used to identify the line that best describes this relationship, used to predict future behaviors

Y=a+bX (a=intercept of y, b=slope)

formula for regression

Pearson correlation (parametric)

perfect relationship between two variables that when graphed resembles a straight line 

regression line

imaginary line or best fitting line, represents linear relationship between two variables 

less

The closer this coefficient is to 0, the (more or less) likely it is that these two variables are correlated to each other

mean

describe one group (parametric)

median

describe one group (nonparametric)

1 sample t test

compare one group to a hypothetical value (parametric)

Wilcoxon test (chi square for two possible outcomes)

compare one group to a hypothetical value (nonparametric)

2 sample unpaired t test

compare two unpaired groups (parametric)

Mann-Whitney test

compare two unpaired groups (nonparametric)

2 sample paired t test

compare two paired groups (parametric)

Wilcoxon test

compare two paired groups (nonparametric)

One-way ANOVA

compare 3 or more unmatched groups (parametric)

Kruskal-Wallis test (or Chi square for two possible outcomes)

compare 3 or more unmatched groups (nonparametric)

linear or nonlinear regression

predict value from measured values (parametric)

nonparametric regression

predict value from measured values (nonparametric)

Spearman’s (nonparametric)

data doesn’t follow a specific probability distribution

the respondent, the interviewer, the instrument, the mode of data collection

Measurement errors are commonly ascribed to four sources:

social desirability

represents what they should ideally do

construct, content, face, criterion

measurement or test validity

internal and external validity

design or experimental validity

selection bias, testing effect, situation effect, experimenter effect, and Hawthorne effect

threats to internal validity

population

entire group of people that are the focus of the study

sample

The procedure used by researchers to select a subset of the population

probability- quantitative, nonprobablity- qualitative

what type of study is probability and nonprobality

every member has an equal chance of participation, and

it ensures representation of the population.

advantages of simple random sampling

requires a complete list of the population from which the sample is drawn

disadvantages of simple random sampling

simple random sampling and stratisfied random sampling

types of probability sampling

convince, snowball, quota, purposive

types of nonprobablity sampling

quota sampling

what kind of sampling is: Researchers conducting a survey

about consumer opinions on a new product,

where they set quotas for different

demographics (age and gender): interviewing

20% young adults, 30% middle-aged, and 50%

seniors, with equal representation of males and

females within each age group.

stratified random sampling

what kind of sampling is: if a class has 20 students, 18 female and 2 male, and a researcher

wanted a sample of 10, the sample will consist of 9 randomly chosen females

and 1 randomly chosen male, to represent this population.

Logical formatting

Take a critical look at your survey questions before beginning your data entry

Determine how the variables can best be organized to ensure that the data

analysis phase is simple

software packages

Common programs for entering and analyzing quantitative data are

Excel

Statistical Package for Social Science (SPSS)

Statistical Analysis System (SAS)

 Stata

Software for statistical computing and graphics (R)

unique identification

Certain ground rules need to be

established before data entry

oFirst assign a unique

identification number to every

participant or case in your

study.

 First column in your

spreadsheet.

coding and codebooks

translating words into numbers, can simplify raw information in a way that is easier to analyze

necessary for running any analysis because of the various analytical requirements of the variables

why is identifying each variable necessary

codebook

key to the codes in your study and how they correlate with participant responses.

o Extremely important document to be kept in a separate and secure file. Info could be lost without this

it can signal an underlying problem with a specific question

why is missing data important for your analysis

row

represents one case or participant

column

represents one variable in the same category for every single case or participant, calls for a descriptive name

1) Allocate Responsibilities among Team Members

2) Create a Detailed Daily Timeline

3) Set Milestones and Check-in Points

steps to creating a 14 day data collection plan