Collecting Data

Limit the survey to those who have listed exercise as an interest on social media profiles.

Only call people during business hours on weekdays to ensure consistency in responses.

Use voluntary response where only interested individuals call back to give their answers.

Follow up with multiple calls at different times of day or days of week if there is no initial response.

Standard deviation

Range

Variance

Interquartile range

Response bias, since respondents may not feel comfortable sharing truthful answers over the phone.

Undercoverage, because it misses segments of the population that do not have landlines.

Sampling variability, as results might differ from sample to sample even if they represent the same population.

Nonresponse, due to people refusing to participate in phone surveys.

Undercoverage bias

Response bias

Nonresponse bias

Voluntary response bias

Undercoverage bias

Nonresponse bias

Response variance

Observer bias

Points forming distinct groups but within each group there’s a clear positive trend between variables.

A plot where points decrease consistently from left to right indicating negative association.

A plot where points form a shapeless cloud without any discernible trend or direction.

Points clustered tightly along a line sloping upwards from left to right indicating positive association.

It can result in overestimating or underestimating treatment effects.

It eliminates the need for blinding participants to treatment allocation.

It ensures equal variability among treatment groups, thus simplifying analysis.

It increases the likelihood that confounding variables will be evenly distributed across treatments.

The sample may not be representative of the population.

The sampling frame might be incomplete.

The standard deviation is biased.

The sample may be too small.

The overall fit of the model is enhanced by their presence.

Influential points confirm that there's equal variance among residuals, indicating homoscedasticity.

The estimated slope and intercept may heavily depend on these points, thus being less reliable.

These influential points indicate strong evidence against any outliers existing in the dataset.

Misuse of p-values in interpreting significance of a relationship instead of assessing effect size relevance in practical scenarios

Low response rates resulting in small affected statistical power reduce the chance of detecting a significant relationship that exists

Confounding variables like socio-economic status aren't controlled during the analysis phases, thus confounding outcome interpretations

Selection bias due to self-selection nature of online surveys leads to potential nonrepresentative samples