Quiz
Inferential Stats Practise Questions
Crie uma conta grátis para terminar este quiz
Salve seu progresso e veja sua nota final.
Já tem uma conta? Inicie sessão
Perguntas do quiz (65)
1How could you tell whether the data collected from JIG:SAW employees and the data from non-employees were meaningfully different?
- AIt's not possible to know with certainty, but only to know the probability that they are different
- BNull hypothesis significance testing
- CIf the observed data allows you to probabilistically rule out the null hypothesis, then you might believe the alternative hypothesis: that the two are meaningfully different
- DAll of the above
2By convention, psychologists generally set the p-value at...
- A10%
- B50%
- Cchance
- D5%
3Null hypothesis testing can tell you...
- AWhether the null hypothesis is true
- BWhether the alternative hypothesis is true
- CWhether the alternative hypothesis is false
- DWhether the alternative hypothesis is likely to be false
4Which of the following represents the test statistic?
- Asignal/noise
- Beffect/error
- Cb/standard error of b
- DAll of the above
5What is another name for the experimental hypothesis?
- AThe null hypothesis
- BH0
- CThe alternative hypothesis
- DBoth A and B
6In order to decide whether the difference between JIG:SAW employees and non-employees were meaningful, Zach needs to use:
- ABetter data than he has
- BA regression analysis
- CNull hypothesis significance testing
- DThe scientific method
7What does it mean when a researcher rejects the null hypothesis at the 0.05 level?
- AThere is a 5% chance os getting an extreme result by chance if the null hypothesis is true
- BThere is a 5% chance of getting such an extreme result by chance if the null hypothesis is false
- CThere is a 5% chance that there is a difference between the null and alternative hypothesis
- DThere is
8The idea of using the p-value to estimate the probability of the null hypothesis, and the idea of testing the null hypothesis against the alternative hypothesis can be attributed to:
- ANeyman and Pearson, respectively
- BPearson
- CPearson and Neyman, respectively
- DNeyman
9What is the alternative hypothesis for the following question: Does watching violent television cause aggressive behaviour?
- APeople who watch violent television programs will behave more aggressively compared to those who dont.
- BPeople who watch violent television will have similar behaviour to those who dont
- CWatching violent television doesn't predict the rate of aggressive behaviour
- DThere will be no difference in the behaviour of people who watch violent compared to those who dont.
10What can you do with a test statistic?
- ACompare it to its probability distribution to see how likely it is to get a test statistic that big if the null hypothesis were true
- BUse it prove the null hypothesis
- CUse it to prove the alternative hypothesis
- DAll of the above
11Which would require a smaller test statistic in order to be considered significant?
- AThere isn't enough information to answer
- BA two-tailed test
- CA one-tailed test
- DThe test statistic regarded as significant would be the same
12A larger sample size is expected to result in ...
- AA smaller standard error of the mean
- BA larger standard deviation
- CA smaller standard deviation
- DA larger standard error of the mean
13The probability that we conclude that there is no effect on our manipulation where, in fact, there is an effect is:
- A20% by convention
- BThe probability of a Type 2 error
- CThe beta level
- DAll of the above
14To test the null hypothesis H0(pop mean = 20) against H1 (pop mean not 20), a random sample of 400 observations is drawn from a population with a st.dev=5. We will reject H0 if our sample mean is greater than 20.75 or less than 19.25. What is our significance level for this experiment?
- AAlpha = 0.05
- BAlpha = 0.0013
- CAlpha = 0.10
- DAlpha = 0.0026
- EAlpha = 0.0052
15Zach sold 62, CI[52,72] albums when he gave away free gifts at his concerts, and 53[50,56] albums when he did not. Are these likely to be statistically significantly different using an alpha level of 0.05?
- AYes
- BNo
16What is true of a Bonferroni correction?
- AIt ensures that the familywise error stays below alpha
- BIt is described by the formula Pcrit = K/a
- CIt cannot be used if there are more than 8 tests being conducted with the same data set.
- DIt is a common way to control for Type 2 errors
17Which of the following set of pairings is correct?
- AType 1; Type 2; power::alpha; beta; 1-alpha
- BType 1; Type 2; power::beta;alpha;1-beta
- CType 1; Type 2; power::alpha;beta;1-beta
- DType 1; Type 2; power::1-alpha;beta;1-beta
- EType 1; Type 2; power::beta;alpha;1-alpha
18Which has more power?
- AThere isnt enough info to answer
- BThey are equal
- CA two-tailed test
- DA one-tailed test
19If you use a value of 0.05 as Fishers p-value, and power = 0.80, what is the (combined) probability that you will make a Type 1 or a Type 2 error?
- A0.025
- B0.05
- C0.20
- D0.25
20According to your textbook's author, you should use a two-tailed test unless:
- AYou would act the same way if you got a null hypothesis as you would if you saw an effect in the non-predicted direction
- BYou would act the same way if you got a null hypothesis as you would if you saw an effect in the predicted direction
- CYou dont have a prediction
- DYou dont think the independent variable will impact the dependent variable.
21Statistical power is affected by all but which of the following?
- AThe size of the effect
- BThe a level
- CSample size
- DThe direction of the alternative hypothesis
22The null distribution has a mean of 7, and the standard error of the alternative distribution is estimated from a sample size of 10. This sample has a mean of 9 and st.dev of 19. Assuming both the distributions are normal, what is the 95% CI of the difference between the means, and what is the result of your hypothesis test based on that CI?
- A-9.77-13.77; fail to reject the null hypothesis because the CI contains zero
- B-2.29-6.29; reject the null hypothesis because the CI doesn't contain the value 7
- C-6.29-2.29; fail to reject the null hypothesis because the CI contains the value zero.
23Without correction, and assuming a value of 0.05 as Fishers p-value, the familywise probability of making a Type 1 error if you test 4 hypotheses with the same data set is
- A0.05
- B0.0975
- C0.143
- D0.1855
24Which study requires a 2-tailed test?
- AYou want to test whether children who skip breakfast have lower test scores than children who eat breakfast
- BYou want to test whether increasing the temperature of the room has an effect on test scores
- CYou want to test whether participating in your weight training program increases an athletes bench press
- DYou want to test whether children who eat breakfast have higher test scores than children who skip breakfast.
25The b-level tells you...
- Athe probability of finding an effect
- Bthe probability that a given test will find an effect when one exists in the population
- Cthe power of the test
- Dthe probability that a given test will not find an effect when one exists in the population
26The best way to increase your power is too...
- AIncrease sample size
- BChoose a stricter a-level
- CChoose a smaller 1-b value
- DChoose a greater b value
27Pcrit = alpha / k
- AIs known as the Bonferroni correction
- BAllows an experimenter to test multiple hypotheses with one data set w/o increasing the probability of a Type 1 error
- CIncludes the variable k which represents the # of hypotheses being tested.
- DAll of the above
28Using the Bonferroni correction, what is the adjusted probability to support an alternative hypothesis with a Type 1 error rate of 3% when 5 tests have been conducted?
- APcrit=0.006
- BPcrit=0.003
- CPcrit=6%
- DPcrit=0.06
29Believing that your independent variable affects your dependent variable:
- Acould make you vulnerable to a Type 2 error
- Bis likely to happen with a probability of 1-a
- Cis likely to happen with a probability of 1-b
- Dcould make you vulnerable to a Type 1 error
30Increasing your sample size will make it more likely that
- AYou can accept the null hypothesis
- BYour finding is meaningful
- CYour finding is important
- DYour finding is significant
31Increasing your sample size will have an effect on all but which of the following?
- AThe standard error of the mean
- BThe relationship between the sample estimates and population parameters they estimate
- Cp-value
- DCohen's d
32After completing your analyses, you find that p > 0.05, your pre-determined alpha value. You can now conclude that
- AThe null hypothesis is true
- BThere is no effect
- CThe effect is not big enough to be found in the sample that we have taken
- DAll of the above
33If you are comparing two groups (e.g. IQ of students in 3XE3 compared to the IQ of all McMaster students), what of the following statistic will you not include in your calculation of cohen's d? (Assume equal standard deviations across groups).
- AThe mean of the control group
- BThe standard deviation of the experimental group
- CThe standard deviation of the control group
- DThe mean of the experimental group
34An example of effect size is
- ACohen's d
- BPearsons Correlation
- CThe odds ratio
- DAll of the above
35Once completed, and assuming a=0.05, the probability of a Type 1 error on an individual experiment is _______, and the probability of a Type 1 error across 1000 identical experiments is _______.
- A0.05, either 1 or 0
- B0.05, 0.05
- CThere isn't enough information to answer
- DEither 1 or 0, 0.05
36Once completed, and assuming b=0.20, the probability of a Type 1 error on an individual experiment is _______, and the probability of a Type 1 error across 1000 identical experiments is _______.
- A0.20, Either 1 or 0
- BEither 1 or 0, 0.20
- CThere isn't enough information to answer this question
- D0.20, 0.20
37Imagine the mean of your contol group (n=1000) was 10 and the standard deviation 6. The mean of your experimental group (n=10) was 15 and the standard deviation 10. Compute cohen's d.
- A0.625
- B0.83
- C0.5
- D1.5
- EThere isn't enough information to make this calculation
38In order to quantify the effect size of an experiment, you compare the mean of one group to the mean of the other group. All of the following will make this effect size hard to compare except:
- AThe mean will be expressed in the units of this study measurements
- BThis figure accounts for the signal, but not the noise
- CThe construct that is being measured is specific to this study
- DNone of the above
39What are the units of measurement of a z-score?
- AStandard deviations
- BVariance
- CMeans
- DStandard errors of the mean
40A problem with NHST that your textbook describes is
- Ain order to use NHST, you have to assume that the null hypothesis is true
- Bit doesn't take into account the prior probability that there is (or isn't) an effect
- Can important effect can be missed because the sample size is too small
- Dthe all-or-nothing nature of the conclusions; an important effect can be missed because the p-value is close to, but not less than 0.05
- EAll of the above
41According to your textbook, how is it that the intentions of scientists influence NHST?
- Apeople are biased and will. consciously or unconsciously, interpret the results as consistent with their expectations
- Bthe number of participants who actually participated in the experiment should determine the p-value, regardless of the number of participants that the experimenter intends to collect data from
- Cthe p-value that is appropriate depends on what your intentions were for collecting data
- DAll of the above
42An effect size is:
- AAn alternative type of significance test
- BA measure of the effect in the population
- CA measure of the effect in the sample
- DAll of the above
43Over many experiments, you are able to consistently demonstrate Effect a (conclude A1 over A0) with a relatively small sample size. In contrast, you need a relatively large sample size to consistently demonstrate Effect B (conclude B1 over B0). What is the best interpretation for this difference between Effect A and Effect B?
- AEffect B results are more statistically significant than Effect A.
- BEffect B results are more practically significant than Effect A
- C(A0-A1) / S is likely larger than (B0-B1)/ S
- D(A0-A1) / S is likely smaller than (B0-B1)/ S
- ENone of the above
44In order to claim you have a large effect, your Cohen's d should be:
- AAt least 0.2
- BAt least 0.5
- CAt least 0.8
- DAt least 0.99
45Which of the following has a maximum of 1:
- ACohen's d
- BPearson's correlation
- CThe odds ratio
- DAll of the above
- EA and B only
46You conduct a significance test across a huge sample. The resulting p-value is less than 0.05. This likely means that...
- Ayour finding is significant
- Byour finding is meaningful
- Cyour finding is important
- Dall of the above
47In Bayes theorem, the prior probability is ...
- Athe probability of finding the data you have collected
- Bhow likely you believe it is that your hypothesis is true, prior to collecting any data
- Cthe probability of finding the data you have collected, given the alternative hypothesis
- Dthe probability of the hypothesis, given the data you have collected
48In Bayes' theorem, the posterior probability is ...
- Athe probability of finding the data you have collected, given the alternative hypothesis
- Bthe same as the null hypothesis in NHST
- CThe probability of the hypothesis, given the data you have collected
- Dequal to the base rate
49In Bayes' theorem, the posterior odds are ...
- AThe posterior probability of one hypothesis divided by the posterior probability of the other hypothesis
- Bhow likely you believe it is that your hypothesis is true, prior to collecting any data
- Cthe probability of finding the data you have collected
- Dthe probability of the hypothesis, given the data you have collected
50Which Baye's factor suggests that you have substantial evidence in favour of the alternative hypothesis?
- AAny value between 1 and 3
- BAny value between 0 and 1
- CAny value between 3 and 10
- DValues greater than 10
51In Bayesian analysis, you use the data you collect to...
- Atest whether the data set was likely to have been observed by chance
- Btest the alternative hypothesis
- Ctest the null hypothesis
- Dupdate your prior beliefs
52A meta-analysis...
- ASummarizes effect sizes across a number of studies
- BAddresses the concern that some published research might report "effects" that result from type 1 errors
- CRequires you to average p values across a number of studies and use that to decide whether to reject the null hypothesis
- DAll of the above
- EA and B only
53In a meta-analysis, you want to give more weight to studies that are more____ and that will be determined by ____.
- Aprecise, sample size
- Bwidely accepted, peer-review
- Caccurate, sample size
- Daccurate, sampling methods
54In Bayes theorem, the probability of the data given the model is:
- Athe likelihood
- BComparable to the question NHST asks
- CAll of the above
- DA and B only
55In the equation (Outcomei = (b0 +b1Xi) + errori) if your predictor variable = 0, what is the best guess for the value of the outcome?
- A+ or - errori
- Bb0
- Cb1
- DThere isnt enough information to answer
56Which of the following is the linear model?
- AAn equation describing the relationship between the concentration of a drug and the outcome
- BAn equation describing the relationship between group membership (got drug v/ got a placebo) and the outcome
- Cb0 +b1Xi
- DAll of the above
57The difference between the value of the outcome predicted by the model and the value of the outcome actually observed is called the ...
- Aerror
- Bresidual
- Cvariance
- DA and B only
58We interested in whether _____ is a good fit to the observed data.
- AOur model, for which parameters have been calculated based on our data
- Bb0 +b1Xi, where b1 is a parameter describing the relationship between our independent variable and the outcome or dependent variable
- Cb0 + b1X1i + b2X2i +... +bnXni
- Dmore generally, an equation that summarizes the pattern of observed data
- EAny of the above are possibly good answers
59The following assumptions are listed in order of importance :
- AIndependent errors, Homoscedasticity
- BAdditivity and linearity, predictors are uncorrelated with external variables, Homoscedasticity.
- CNo perfect multicollinearity, Homoscedasticity
- DPredictors are uncorrelated with external variables, Homoscedasticity
60With the equation (Outcomei = (b0 + b1Xi) + errori)...
- Athe parameters are estimated based on your collected data
- Bfor each parameter, you can calculate a test statistic
- Cfor each parameter, you can calculate a p-value
- DAll of the above
61Which of the following is NOT true about the Assumption of independent errors?
- AIts the most important of all the assumptions of the general linear model
- BIf it's violated, you'll want to use a multilevel linear model, which can take into account the relatedness of the errors
- CIt can be tested using the Durbin-Watson test, but only if the order in which data points are collected is meaningful
- DIs important because the estimate of the standard error relies on observations being independent, and the standard error is used to compute confidence intervals and test statistics.
62In a relationship between the predictor variable and the outcome variable can be best described by a straight line, you have satisfied which of the following?
- Alinearity
- Bno perfect multicollinearity
- Cpredictors are uncorrelated with external variables
- Dhomoscedasticity
63In the equation Outcome = (b0 + baXi) + errori, how do you know the value of b1?
- Awe measure it in the population
- Bwe measure it in the sample
- Cwe use the sample data to estimate the alues of the paramters in the models
- Dany of the above are possibly good answers
64Which of the following is false?
- AA model is always imperfect
- BYou use a statistical model to predict the outcome for a given case
- CA model is always linear
- DA model is derived from a specific data set
65In the equation Outcome = (b0 + b1Xi) + error is related to nearly everything we will discuss in this class except:
- ABayes theorem
- BANOVAs
- CChi-square tests
- DThe ttest
Informações
- Descrição
- Mathematics Quiz sobre Inferential Stats Practise Questions , criado por Asees Dhinsa em 27-11-2019.
- Autor
- Asees Dhinsa
- Idioma
- English
- Visualizações
- 1
- Etiquetas
- Avaliação
- Ainda sem avaliações Avalie este quiz