Quiz
COMM 04 Week 11.2 Conclusion Validity and Hypothesis Testing
This quiz has question types that open in the usual player.
Questions in this quiz (23)
1([blank_start]Statistical[blank_end]) [blank_start]Conclusion[blank_end] Validity ---> ([blank_start]Statistical[blank_end]) [blank_start]Analysis[blank_end] ---> Did you come to the right ([blank_start]statistical[blank_end]) conclusion?
- AStatistical
- BConclusion
- CStatistical
- DAnalysis
- Estatistical
2[blank_start]Conclusion[blank_end] Validity: the degree to which conclusions we reach about relationships/differences in our data are reasonable.
When you investigate a relationship there are (in general) [blank_start]two[blank_end] possible conclusions you can reach:
(A) There [blank_start]is[blank_end] a relationship
(B) There [blank_start]is not[blank_end] a relationship
You will decide one way or the other, based on the data you have. [blank_start]Conclusion[blank_end] validity is an assessment of how [blank_start]likely[blank_end] it is that you made the correct decision.- AConclusion
- Btwo
- Cis
- Dis not
- EConclusion
- Flikely
3[blank_start]Hypothesis[blank_end]: A testable prediction about what is or is not true about a particular phenomenon.
Generally states a [blank_start]relationship[blank_end] between [blank_start]independent[blank_end] and [blank_start]dependent[blank_end] variables or a [blank_start]difference[blank_end] in a [blank_start]dependent[blank_end] variable between levels of an [blank_start]independent[blank_end] variable.
[blank_start]Relationship[blank_end]: The correspondence between two variables.
[blank_start]Difference[blank_end]: The degree to which one case or group of cases are (dis)similar from another case or group of cases.- AHypothesis
- Bindependent
- Cdependent
- Drelationship
- Edifference
- Fdependent
- Gindependent
- HRelationship
- IDifference
4[blank_start]Directionality[blank_end] of Hypotheses
[blank_start]One[blank_end]-tailed: Predicts the specific nature of the relationship or difference
• There is a [blank_start]positive[blank_end] relationship between Variable X and Variable Y
• Group A has a [blank_start]higher[blank_end] mean on Variable Y than Group B
[blank_start]Two[blank_end]-tailed: Predicts only that there is a relationship or difference
• There [blank_start]is a relationship[blank_end] between Variable X and Variable Y
• Group A has a [blank_start]different[blank_end] mean on Variable Y than Group B
NOTE: The choice of [blank_start]one[blank_end]-tailed or [blank_start]two[blank_end]-tailed should be based on [blank_start]theory[blank_end]*
In practice, many researchers are more comfortable using [blank_start]two[blank_end]-tailed hypothesis tests, but that is a separate consideration (and one we will come back to shortly).- ADirectionality
- BOne
- CTwo
- Dpositive
- Eis a relationship
- Fhigher
- Gdifferent
- Hone
- Itwo
- Jtheory
- Ktwo
5[blank_start]Research[blank_end] Question: A question about what is or is not ‘true’ about a
particular phenomenon. “Explicit question researchers ask about variables of interest.”
The decision about whether to frame your tests as a [blank_start]research[blank_end] question vs. a [blank_start]hypothesis[blank_end] is based on whether there is sufficient [blank_start]evidence[blank_end] available from which to make a prediction.- AResearch
- Bresearch
- Chypothesis
- Devidence
6Directionality
• In principle: You can pose both directional (cf. [blank_start]one[blank_end]-tailed) and non-
directional (cf. [blank_start]two[blank_end]-tailed) research questions.
• In practice: If you have enough information to pose a [blank_start]directional[blank_end] research question, you have enough [blank_start]information[blank_end] to make a prediction- Aone
- Btwo
- Cdirectional
- Dinformation
7Types of Hypothesis: [blank_start]Null[blank_end] v. [blank_start]Alternative[blank_end]
[blank_start]Null[blank_end] Hypothesis: The prediction that there is no difference/no relationship
• We assume the [blank_start]null hypothesis[blank_end] is true until we obtain evidence that it is likely false i.e., we find support for the alternative hypothesis
• Symbolized as H0 (that’s an italicized H followed by a subscripted zero)
[blank_start]Alternative[blank_end] Hypothesis—The logical alternative to the null hypothesis
• Also referred to as the researcher's hypothesis; this is the prediction that the researcher makes
• Symbolized as H, Ha, or H1 (often there are multiple hypotheses, so H 1, H 2 , H 3 , etc.)
• For research questions, this is changed to RQ, RQ a, or RQ 1 in the research report- ANull
- BAlternative
- CNull
- Dnull hypothesis
- EAlternative
8Null Hypothesis—The prediction that there is no difference/no relationship
• We assume the null hypothesis is [blank_start]true[blank_end] until we obtain evidence that it is likely [blank_start]false[blank_end] i.e., we find support for the alternative hypothesis
• Symbolized as [blank_start]H0[blank_end] (that’s an italicized H followed by a [blank_start]subscripted zero[blank_end])
Alternative Hypothesis—The logical alternative to the null hypothesis
• Written such that if the null hypothesis [blank_start]is TRUE[blank_end], then the alternative hypothesis [blank_start]must be FALSE[blank_end]
• If [blank_start]H0 is FALSE[blank_end], then Ha [blank_start]must be TRUE[blank_end]. For example, if:
Ha: Group A has a higher mean than Group B on Variable Y (μA > μB), then
H0 : Group A has a mean that is less than or equal to Group B on Variable Y (μA ≤ μB)- Atrue
- Bfalse
- CH0
- Dsubscripted zero
- Eis TRUE
- Fmust be FALSE
- GH0 is FALSE
- Hmust be TRUE
9[blank_start]Alternative[blank_end] Hypothesis—The logical alternative to the null hypothesis
• We can never prove Ha, we can only show that H0 is [blank_start]unlikely[blank_end], given a set of observations
• If we find that the null hypothesis is not likely to be true given the observed data, then we (a) [blank_start]reject[blank_end] the null hypothesis, and (b) conclude that the alternative hypothesis is [blank_start]supported[blank_end]- AAlternative
- Bunlikely
- Creject
- Dsupported
10The Basic Hypothesis Testing Steps
1. Set a [blank_start]confidence level[blank_end].
2. Conduct an [blank_start]appropriate statistical[blank_end] test.
3. Compare [blank_start]calculated[blank_end] value to [blank_start]critical[blank_end] value.
• [blank_start]Calculated value[blank_end]—The test statistic that you compute
• [blank_start]Critical value[blank_end]—Value of the random variable at the boundary for accepting or rejecting the null hypothesis.
• The value that the “calculated value” must be [blank_start]greater[blank_end] than (in absolute terms) to achieve statistical significance at a specific confidence level
• If the calculated test statistic is > critical value of the test statistic, then the test is “statistically significant.” We (a) [blank_start]reject[blank_end] the null hypothesis, and (b) conclude that the alternative hypothesis is [blank_start]supported[blank_end].- Aconfidence level
- Bappropriate statistical
- Ccalculated
- Dcritical
- ECalculated value
- FCritical value
- Ggreater
- Hreject
- Isupported
11[blank_start]Testing[blank_end] Hypotheses: [blank_start]Z-Test[blank_end]
[blank_start]Two sample[blank_end] z-Test: Compares two sample means to each other to
determine if they are drawn from the same population or not
• Research question: Is there a difference in teaching effectiveness [operationalized as exam scores] between live teaching and mediated teaching?
• COMM 304 students randomly assigned to
(A) Live instruction,
or (B) Mediated instruction
• Do the instruction methods really differ from one another? (i.e.,
does the medium of instruction make a difference in terms of test
scores?)
Hypothetical study:
𝑋𝑋Live = 42
𝑋𝑋Mediated = 38
𝑋𝑋L − 𝑋𝑋M = 42 − 38 = 4- ATesting
- BZ-Test
- CTwo sample
12Types of Hypotheses
[blank_start]ISSUE[blank_end]: Are observations caused by some source (e.g., the manipulation) or by random [sampling/assignment] error?
• E.g., what is the probability of randomly selecting two samples from the same population with a magnitude of difference at least as large as the observed difference in sample means (e.g., �𝑋𝑋L − �𝑋𝑋M = 4)?
• Ha : μ1 ≠ μ2
• The difference in teaching methods is not attributable to chance (i.e., there is a statistically significant difference between the means; the two samples represent distinct populations)
[blank_start]Research[blank_end]/[blank_start]Alternative[blank_end] Hypothesis: Statement that the difference is not due to random [sampling/assignment] error
[blank_start]Null[blank_end] Hypothesis: Statement that the difference is due to random error
• H0 : μ1 = μ2- AISSUE
- BAlternative
- CResearch
- DNull
13Understanding Error
[blank_start]Effect[blank_end] size: The strength of a relationship or the magnitude of the difference.
- AEffect
14We can never [blank_start]prove[blank_end] (nor [blank_start]disprove[blank_end]) a hypothesis using a statistical test.
All we can say is the probability of observing a calculated effect size (and direction) given the null hypothesis (H0 ) is [blank_start]true[blank_end].
This is what we refer to as the [blank_start]p-value[blank_end].- Aprove
- Bdisprove
- Ctrue
- Dp-value
15Where statistical conclusions are always probabilistic, there is an inherent amount of [blank_start]error[blank_end] in the conclusions we make.
Type [blank_start]I[blank_end] Error—When a researcher [blank_start]rejects[blank_end] the null hypothesis (there is an effect), when the null hypothesis should be accepted (there is no effect). AKA: A false [blank_start]positive[blank_end]
Type [blank_start]II[blank_end] Error—When a researcher [blank_start]accepts[blank_end] the null hypothesis (there is no effect), when the null hypothesis should be rejected (there is an effect). AKA: A false [blank_start]negative[blank_end]- Aerror
- BI
- CII
- Drejects
- Epositive
- Fnegative
- Gaccepts
16The Type [blank_start]I[blank_end] (or α) error rate is set by the researcher. It is the cut-off point between [blank_start]accepting[blank_end] and [blank_start]rejecting[blank_end] the [blank_start]null[blank_end], e.g., p < .05
If we want less Type I error (a larger confidence interval) then we [blank_start]change[blank_end] α.- AI
- Bchange
- Caccepting
- Drejecting
- Enull
17The Type [blank_start]II[blank_end] (or β) error rate is a function of the [blank_start]true[blank_end] effect size and the research design.
All else equal, we can [blank_start]reduce[blank_end] Type II error (increase power) by (a) i[blank_start]ncreasing[blank_end] sample size or (b) using a more powerful test (e.g., use a [blank_start]parametric[blank_end] test instead of a [blank_start]non-parametric[blank_end] test; use a [blank_start]one[blank_end]-tailed test instead of a [blank_start]two[blank_end]-tailed test)- AII
- Btrue
- Creduce
- Dncreasing
- Eparametric
- Fnon-parametric
- Gone
- Htwo
18Preview: Key Characteristics of the [blank_start]Normal[blank_end] Curve
• Most of the scores cluster in the [blank_start]middle[blank_end] of the distribution
• The [blank_start]normal curve[blank_end] is perfectly [blank_start]balanced[blank_end] or [blank_start]symmetrical[blank_end]
• The tails will never touch the [blank_start]x-axis[blank_end]
• All three measures of central tendency fall at the [blank_start]same[blank_end] point (the center)
• The normal curve has a [blank_start]constant[blank_end] relationship with the standard deviation
With a random sample, the single best estimate of μ is �𝑋𝑋 (see: Central Limits Theorem)
It is not a perfect estimate, however, we can use the sampling distribution to calculate a confidence interval for the estimate.- ANormal
- Bmiddle
- Cnormal curve
- Dbalanced
- Esymmetrical
- Fx-axis
- Gsame
- Hconstant
19Back to the z-test
Conducting a Two-Sample z-Test
Two sample z-tests use Sampling [blank_start]Distribution[blank_end] of [blank_start]Sample Mean Differences[blank_end]
AKA, the “[blank_start]distribution[blank_end] of differences”; one version of the normal distribution
𝑧 = ( 𝑋𝑋1 − 𝑋𝑋2) − 0 divided by σ diff- ADistribution
- BSample Mean Differences
- Cdistribution
20Critical [blank_start]value of z[blank_end], for a [blank_start]two[blank_end]-tailed test, at p < .05 = ±[blank_start]1.96[blank_end]
- Atwo
- B1.96
- Cvalue of z
21X Live-42
X Mediated=38
σ diff = 1.16
z=(X1-X2)-0 divided by σ diff= +[blank_start]3.45[blank_end]- A3.45
22What if this had been a one-tailed hypothesis?
Critical value of z, for a one-tailed test, at p < .05 = ±[blank_start]1.64[blank_end]
Critical value of z, for a one-tailed test, at p < .01 = ±[blank_start]2.33[blank_end]
Ha : Teaching effectiveness [as operationalized by exam scores] will be greater in the live teaching condition than in the mediated teaching condition.
𝑧 = (42 − 38) − 0 divided by 1.64 = +[blank_start]𝟑. 𝟒𝟒𝟒𝟒[blank_end]
[blank_start]Reject[blank_end] H0 at p < .01. Teaching effectiveness is [blank_start]greater[blank_end] in the live teaching condition
than in the mediated teaching condition (z = 3.45, SE = 1.16, p < .01).- A1.64
- B2.33
- C𝟑. 𝟒𝟒𝟒𝟒
- DReject
- Egreater
23What if the direction of they hypothesis were reversed?
Critical value of z, for a one-tailed test, at p < .05 = ±[blank_start]1.64[blank_end]
Critical value of z, for a one-tailed test, at p < .01 = ±[blank_start]2.33[blank_end]
Ha : Teaching effectiveness [as operationalized by exam scores] will be [blank_start]lesser[blank_end] in the live teaching condition than in the mediated teaching condition.
𝑧 = (42 − 38) − 0 divided by 1.16 = +𝟑. 𝟒𝟒𝟒𝟒
[blank_start]Accept[blank_end] H0 at p < .01. Teaching effectiveness is not lesser in the live teaching
condition than in the mediated teaching condition (z = 3.45, SE = 1.16, p > .05).
Hypothesis Notation (two-sample)
Two-tailed
H0 : μ1 = μ2; Ha : μ1 ≠ μ2
One-tailed
H0 : μ1 ≥ μ2; Ha : μ1 < μ2
H0 : μ1 ≤ μ2; Ha : μ1 > μ2- A1.64
- B2.33
- Clesser
- DAccept
Information
- Description
- Undergrad COMM 304 Unit 2 Exam Quiz on COMM 04 Week 11.2 Conclusion Validity and Hypothesis Testing, created by Sarah Kassaye on 14/11/2023.
- Author
- Sarah Kassaye
- Language
- English
- Views
- 3
- Tags
- Rating
- No ratings yet Rate this quiz
More on Types of Validity and Threats to Validity