Week 2/3- More Principles, sampling, measurement and hypotheses Flashcards
What is validity?
when we are successful in measuring what we want to measure, we will have no measurement error
what is random error?
When external factors influence a respondents response to a question
can you do much to prevent random error?
no
what is systematic error?
the way we ask our questions influences the respondents answer eg leading question
can you do much to prevent systematic error?
yes
what is test- retest reliability?
measuring something twice to make sure the outcomes are the same. demonstrates the effectiveness of the random sampling
how does the population of interest affect validity?
You need to ask questions to the right people
what is probability sampling?
every individual in the population has an equal chance of being included in the sample. Also called random sampling
what is multi-cluster sampling?
when you split the population into strata and then randomly select various strata from within the initial strata
what are the benefits of multi-cluster sampling?
reducing costs and reducing travel time
do larger sample sizes provide more accurate findings than smaller ones and why?
yes and smaller sample sizes have a greater probability of including outliers
when our sample size gets larger, what happens to our confidence interval?
it gets smaller
what is a type II error as a result of a small sample size?
where a false null hypothesis is rejected
what are the two implications of a smaller sample size?
1) decrease in statistical power: Increases the probability of type II errors
2) Missing values: not all respondents reply to all of the questions
what is a confidence interval?
a plus or minus figure indicating the range that an estimated statistic probably falls between. Also called margin of error
what is the confidence level?
tells you how sure you can be that your findings would be repeated if you did the survey again
what is a coverage error?
occurs when the list from which sample members are drawn from does not accurately represent the population that you are interested in
what is a sampling error?
the difference between the estimate produced when only a sample of the population is surveyed and the estimate produced when everyone in a population is surveyed
what is a non-response error?
the difference between the estimate produced when only some of the sample respond compared to when everyone in the sample responds. eg some people didn’t participate because the survey was online etc
what is a measurement error?
the difference between the estimate produced and the true value because respondents gave inaccurate answers to the survey questions. This can happen due to social desirability bias.
what is a measurement error?
the difference between the estimate produced and the true value because respondents gave inaccurate answers to the survey questions. This can happen due to social desirability bias.
what is a type I error?
Detecting a non-existent effect
what are descriptive statistics?
used to describe the characteristics of data- tend to be uni-variable (one variable)
what are analytical statistics?
describe the relationship between different variables- tend to be bi-variate or multi-variate
what is the trimmed mean?
the same as the mean but the top 5% results and the bottom 5% results are not included
What type of statistics does the measure of centrality belong too?
Descriptive
give examples of analytic categories
age and height
demographics and purchasing behaviour
job satisfaction
Customer satisfaction and repeat purchasing
what are correlation variables?
the variables are related but we cannot say that one causes the other
what are causation variables?
the variables are related and changes in one variable causes changes in the other
what are the four necessary pre-conditions for causality?
1) Temporal precedence: the presumed cause must occur before the presumed effect
2) Association: Variation in the presumed cause must be related to that in the presumed effect
3) Isolation: No other explanations of causality between two variables
4) The direction of the causal relationship is correctly specified
why is the dependant variable called the dependant variable?
Because it is dependant on the result of the independent variable
What is standard deviation?
How much the results were spread out around the mean result
what are the four steps in hypothesis testing?
1) Specify the null hypothesis in a statement of no effect or relationship
2) Specify the critical value ( the p-value blow which we reject the null hypothesis) - this is usually <0.05
3) Decide on the most appropriate test to use and run that test
4) Use the resultant p-value to reject or accept the null hypothesis
What are the for outcomes to a hypothesis test?
1) There really is no relationship present and that’s what our test found
2) There is no relationship present but our test says there is one- Type I error
3) An effect is present and our test supports this
4) An effect is present but our test suggests there isn’t- Type II error
how can we reduce the probability of both Type I and II errors?
Increasing the power of our statistical test